Sunday, February 23, 2014

The Medicaid Emergency Department Study

There is an important study on the emergency department (ED) and health care policy in the January 17 edition of the journal Science. It looks at the question of whether not health insurance increases or decreases ED use.  This has been a political football for years.  The debate has been that increased insurance enrollment would prevent excessive ED utilization but the evidence has been sparse.  Some surveys have shown that the uninsured view the high cost of ED services and the financial repercussions are a deterrent.  On the increased utilization side is the economic argument that prepaid services lower the cost and therefore increase the use of all medical services across the board.  Another variable is the overall economy.  In an economic downturn, people use less goods and services including medical services.

Mapped onto the ED utilization problem is the EMTLA law or The Emergency Medical Treatment and Labor Act.  This law states that no person requiring medical stabilization can be turned away from an ED based on ability to pay.  A variety of mechanisms shifts the cost of care to the facility and physicians providing the care.  In the case of psychiatric services, EDs are obligated to find an open bed to transfer the patient.  In most states the majority of hospitals with EDs do not have psychiatric units, and that can result in patients being held for long periods of time until a bed opens up or transfer to beds across the state.  More radical solutions to that problem have included discharging a person untreated back out to the street or discharging them after a certain time interval if a bed could not be identified.

The scope of the problem of psychiatric services in the ED has not been well studied.  Some of the large studies suffer from an inadequate look at diagnoses, crisis care, patient flow and disposition and outcomes.  Before this study, I could not find any studies with adequate detail about diagnoses.  The other consideration is selection bias.  In most metropolitan areas, emergency services brings patients with psychiatric crises to identified hospitals with the largest psychiatric services.  These services typically have large capacity and become catchment areas for large areas of the states they are located in.  They can also be overwhelmed due to various factors that affect patient flow.  Most of these factors are directly related to the closure and rationing of psychiatric services in acute care but also residential facilities, clinics and community support services for the severely disabled.

The design of this study is interesting because it is randomized based on a political initiative.  In 2008, Oregon started a limited expansion of Medicaid.  They drew 30,000 names from a pool of 90,000 people.  There were 8 drawings between March and September 2008.  Previous studies on outcomes by the same authors showed that Medicaid assignment led to reduce depression and improved general health but it did not impact several general measures of general health, employment, or earnings.  In this study they looked at 12 hospitals that are the catchment area for Portland and surrounding suburbs.  These hospitals have half of the annual admissions in the state.  The study ran for 18 months, and was an intent-to-treat analysis of the randomly selected Medicaid enrollees and the non-selected matched on demographic variables.

The primary result of the study showed that Medicaid enrollment was associated with a significant use in ED services.  The increase was 41% relative to the control group.  There was no difference in the number of visits resulting in admission but increases in most other types of visits, including those that would be treatable in an outpatient clinic.  For some reason these differences were detected in administrative but not self reported data.  The authors look at three potential reasons for those differences.  The discussion of study limitations focuses primarily on the fact that the low income population studied may differ significantly from other low income populations and limit its generalizability.  The author's also comment on how establishing primary care can logically increase the likelihood of ED utilization.  The commonest scenario there is a patient with with either risk factors or chronic illnesses that calls their primary care clinic and is advised to go to the ED because of the anticipated length or complexity of the required evaluation.   That factor could not be studied with the available data.  In the case of psychiatric services that is typically a change in mental status, suicide risk , aggressive behavior or need for intoxication or detoxification.

One of the features of this study of interest to psychiatrists is the supplementary data.  Table S10 lists "Select Conditions (control sample only)" for a total of 17,498 ED visits separated by category.  A total of 1346 or 8.4% of all visits were for "Substance abuse and mental health issues."  Of that sample, 3% were mood disorders, 2% alcohol related disorders, 1.5 % anxiety disorders, 0.9% schizophrenia and psychotic disorders, and 0.8% substance related disorders.  In looking at visits per condition increased ED utilization occurred for injuries, headaches, and chronic conditions but not mood disorders or substance use or mental health disorders.  It is not possible to see the distribution of ED visits by hospital and with what is known about these distributions on metro areas it is likely that a few of the 12 hospitals had most of these visits.

In the weeks to come, I anticipate that there will be an active debate on the economic and political implications of this study.  From a psychiatric perspective it does not really capture the scope of the problem of how we got to the current predicament of discharging people with psychiatric and substance use problems untreated from emergency departments.  Nobody seems to consider that the ED problem exists as a result of rationing at multiple levels and a physician productivity model that values a stereotypical low to moderate complexity visit.  Most clinics and even urgent care settings have limited flexibility to assess some of the suggested ED problems like new chest pain even though in this study 93.1% of the chest pain assessed was nonspecific and 3.5% represented an acute myocardial infarction.  A few conclusions that I come to:

1.  This study is well done, unique and seems to have a highly significant finding that increased insurance to a low income population leads to increased ED utilization rather than less.  Caution is needed in the interpretation of that data.  A major weakness of any study like this is the fact that it is all of the data is administrative rather than clinical.  This is a major weakness of practically every data set used to establish health care policy in the past starting with the RAND studies on overutilization of hospitalizations and procedures relative to what was determined by the PROs of the 1990s.  These studies showed that when the data was reviewed by non-biased reviewers with no conflict of interest, there was minimal to no overutilization.  It is probably time to consider that we need better data.

2.  All elements in the system are not equivalent - no 2 EDs are the same.  In any state you can walk into an ED attached to a Level 1 trauma center and burn unit or one that is staffed by moonlighting physicians or residents who may not be emergency medicine specialists.  That will naturally affect referral patterns and overcrowding phenomenon.  Detailed patient flow pattern in and out of the busiest EDs with enough granular data about that phenomena is probably more important in addressing the problem than a look at a single global insurance decision.  Data in this study and others suggests that the increased ED use is based on rational decision making about medical conditions and previous surveys on wanting to avert a financial catastrophe.

3.  Targeted interventions to reduce ED use is specific populations are highly effective.  Assertive Community Treatment (ACT) teams for people with chronic mental illness are a good example.  In these interventions teams have their own crisis programs independent of EDs as well as medical staff who are available to the patients 24/7.  Their goal is also to avoid psychiatric hospitalizations and they are very good at that.  As clinics are acquired and consolidated under various managed care organizations the likelihood of consulting with a person from your primary care clinic after hours decreases significantly and that probably means more contact with the ED.

4.  Urgent Care facilities are a logical extension of primary care clinics after hours and there is currently no psychiatric equivalent.  A clinic with adequate multidisciplinary mental health staff would seem like a better options than being seen in an ED.  There currently do not appear to be any facilities like this for mental health other than county government based crisis lines that vary considerably form county to county.

Despite all of the considerations I have listed above and more, I do not expect a more sophisticated look at this issue.  Our politicians are incapable of it and the conflicts of interest related to the business side of medicine will typically carry the day.  There will be some ideological arguments about economic theory but in the end, what is good for business will carry the day.

Increased utilization of the ED is looking better and better for business every day.    


George Dawson, MD, DFAPA

1: Taubman SL, Allen HL, Wright BJ, Baicker K, Finkelstein AN. Medicaid increases emergency-department use: evidence from Oregon's Health Insurance Experiment. Science. 2014 Jan 17;343(6168):263-8. doi: 10.1126/science.1246183. Epub 2014 Jan 2. PubMed PMID: 24385603.

2: Fisman R. Health care policy. Straining emergency rooms by expanding health insurance. Science. 2014 Jan 17;343(6168):252-3. doi: 10.1126/science.1249341.  Epub 2014 Jan 2. PubMed PMID: 24385605.

1 comment:

  1. Check out the Crisis Response Center, Tucson, Arizona. This is exactly what you are talking about in terms of an urgent care center for psychiatry.